KR100979335B1 - Method for automatic quantification of air trapping on chest CT data - Google Patents

Abstract

The present invention comprises the steps of: (a) acquiring three-dimensional minimum breathing chest volume data and three-dimensional maximum breathing chest volume data through CT (Computed Tomography) technique; (b) dividing pulmonary parenchyma and pulmonary vessel from each volume data, and generating a binary image for each volume data; (c) correcting global movement between binary images for each of the volume data by using an affine matching technique for minimizing the distance between the closed surfaces of the minimum and maximum breaths; (d) correcting local motion between the binary images using a regional deformation model based non-rigid registration technique; (e) generating a subtracted image between the binary images whose correspondence is identified through the global and local motion correction; And (f) analyzing and quantifying the subtracted image, thereby providing an automatic air capture quantification method using a chest CT image.

The automatic air capture quantification method uses affine matching and local deformation model based non-rigid image matching to determine the correspondence relationship between minimum and maximum breathing images, and automatically captures the air capture area using image subtraction and analysis. As a result of the analysis, it is possible to automate qualitative reading methods that are traditionally manual. In addition, it is possible to quantitatively diagnose the air trapping area of the lung disease patient, thereby providing quantified data on how much air is trapped in which part of the lung.

Description

Method for automatic quantification of air trapping on chest CT data}

The present invention relates to an automatic air capture quantification method using chest CT images, and more particularly, to an automatic air capture quantification method using minimum / maximum breathing CT images of lung disease patients.

The prevalence of chronic obstructive pulmonary disease, which occurs relatively frequently in Korea, was reported to be 8.7% in the 2002 National Survey. However, the prevalence rate is expected to continue to increase because the smoking population, which is the main cause, is still large.

The major lesion causing airflow obstruction in patients with chronic obstructive pulmonary disease is the peripheral airway. Here, the evaluation of lesions using CT images is very limited, and it can be said that it is limited to the evaluation of air capture due to early obstruction of peripheral airway during respiration. The extent of air capture seen on CT images is related to the degree of inflammatory infiltration in the smooth muscle layer of the peripheral airways. The discovery of these air traps may assist in the early detection of chronic obstructive pulmonary disease, so accurate analysis is of great clinical importance.

Recently, CT (computed tomography) technology has been widely used as an imaging test for diagnosing such air trapping. The main feature of CT imaging for the diagnosis of air trapping is to compare the brightness of the lung parenchyma in the maximal breathing image with the brightness of the same position in the minimal breathing image, which is the area where the change is small. It is to identify the existence and distribution of the area. However, since the regional deformation of the lung is very large according to the breathing of the patient, it is difficult and time consuming to identify the correspondence relationship. Further, in the diagnosis and evaluation, only the doctors are qualitatively evaluated visually, so that the reading results between the doctors may be different from each other, and there is a problem that there is a risk of misdiagnosis.

Unlike the grasping of the correspondence relationship between the minimum and maximum breathing images and the air capture region analysis, which have been performed qualitatively by hand, the present invention uses the affine matching technique and the regional deformation model based non-rigid image matching technique. The purpose of the present invention is to provide an automatic air capture quantification method using chest CT images which grasps the correspondence relationship between the maximum breathing images and automatically analyzes the air capture area using image subtraction and analysis techniques.

The present invention includes the steps of: (a) acquiring three-dimensional minimum breathing chest volume data and three-dimensional maximum breathing chest volume data through CT techniques; (b) dividing pulmonary parenchyma and pulmonary vessel from each volume data, and generating a binary image for each volume data; (c) correcting global movement between binary images for each of the volume data by using an affine matching technique for minimizing the distance between the closed surfaces of the minimum and maximum breaths; (d) correcting local motion between the binary images using a regional deformation model based non-rigid registration technique; (e) generating a subtracted image between the binary images whose correspondence is identified through the global and local motion correction; And (f) analyzing and quantifying the subtracted image, thereby providing an automatic air capture quantification method using a chest CT image.

The present invention also provides a computer-readable recording medium having recorded thereon a program for executing an automatic air capture quantification method using the chest CT image on a computer.

Automatic air capture quantification method using chest CT image according to the present invention, grasp the relationship between the minimum / maximum breathing images using the affine matching method and regional deformation model-based non-rigid image matching, image subtraction and analysis Automatic analysis of air capture zones can be used to automate qualitative reading methods that are traditionally manual. In addition, it is possible to quantitatively diagnose the air trapping area of the lung disease patient, thereby providing quantified data on how much air is trapped in which part of the lung. Furthermore, when visually expressing the air trapping area of a lung disease patient, it is possible to reduce errors in diagnosis and evaluation, as well as to be applicable as a diagnostic aid immediately applicable to actual clinical practice.

1 is a flowchart of an automatic air capture quantification method using chest CT images according to an embodiment of the present invention. 2 is another flow chart of FIG. 1, and FIG. 3 is a system diagram for the method of FIG.

First, prior to the description of the automatic air capture quantification method using the chest CT image, the automatic air capture quantification system 100 for the method will be described as follows. The automatic air capture quantification system 100 includes a photographing unit 110, a display unit 130, an input unit 120, and a control analysis unit 140.

The photographing unit 110 acquires the photographing information by capturing three-dimensional chest volume data of the minimum breath and the maximum breath by a CT technique. The control analyzer 140 divides the lung parenchyma and the pulmonary blood vessels from the obtained photographing information, generates binary images for minimum and maximum breathing, and compensates for global and local motions for each binary image. . In addition, the control analysis unit 140 generates and quantifies the subtracted image between the motion-compensated binary images, and provides quantified data on how much air is trapped in which part of the lung. Of course, in addition to the control analysis unit 140 may control the respective components, such as the photographing unit 110, the display unit 130, the input unit 120.

Meanwhile, the input unit 120 may receive various manipulation signals from a user (expert or doctor) and transmit them to the control analysis unit 140. The display unit 130 may visualize and display the subtracted image in addition to the photographing information. In addition, the control analysis unit 140 processes the signal input to the input unit 120 and transmits a corresponding operation to the display unit 130, so that the corresponding information is displayed on the real-time screen.

Hereinafter, an automatic air capture quantification method using a chest CT image according to an embodiment of the present invention will be described in more detail with reference to FIGS. 1 to 3.

First, the photographing unit 110 obtains the 3D minimum breathing chest volume data and the 3D maximum breathing chest volume data through CT techniques (S110). Each volume data thus obtained is transferred to the control analyzer 140 and processed. In this case, each of the obtained volume data may be, in addition to the data photographed by the photographing unit 110, volume data photographed from the outside and received through the input unit 130.

Next, the control analyzer 140 divides the lung parenchyma and the pulmonary blood vessel from each volume data, and generates a binary image for each volume data (S120). In order to divide the parenchyma and pulmonary blood vessel, seed point region growth method is used as follows.

First, from each volume data image, a background outside the chest having a brightness value similar to the lung parenchyma is removed by performing a two-dimensional seed point region growth method for each slice. At this time, the outermost points of each slice are used as seed points.

In the volume data of the maximum breathing, a threshold value of -1024 HU or more and -400 HU or less is used, and in the volume data of the minimum breathing, a threshold value of -1024 HU or more and -200 HU or less is used to generate a three-dimensional seed point region growth method. To extract the lung parenchyma. The seed point used in the three-dimensional seed point region growth method uses a center point after automatically extracting the bronchial region at the position where the bronchus branches as a template. In this case, when the 3D seed point region growth method is performed using the same seed point at a threshold of -1024 HU or more and -950 HU or less, the bronchus is extracted, and when the bronchial region is removed from the previously extracted lung parenchyma, only the lung remains. Since the seed point growth method is disclosed in the past, a detailed description thereof will be omitted.

In addition, when hole filling is performed in the remaining lung parenchyma, the lung parenchyma containing the pulmonary blood vessels is divided. In this region, more than -400 HU are pulmonary blood vessels, and the remaining regions are pulmonary parenchyma. Here, referring to Equation 1, the lung parenchyma is assigned a value of '0' as a class 1, and the value of '1' as a class 2 for the pulmonary blood vessel and the outside of the lung, respectively, and the binary An image can be generated.

[Equation 1]

Refer to the example image of FIG. 4 for the appearance of each binary image regarding the 3D minimum and maximum breath volume data. After the generation of the binary image, the lung boundary may be extracted (S121). Subsequent registration processes (S130 ˜ S140) are performed on the binary images classified by class, and may be more precisely matched by correcting the difference between the brightness values of pulmonary parenchyma and pulmonary blood vessel between the two binary images.

That is, after the binary image generation step (S120), the control analysis unit 140, using an affine matching technique to minimize the distance between the closed surface of the minimum breath and the maximum breath, to each of the volume data Compensate for the global motion between the binary images (S130). First, a chamfer distance map of the closed surface of the volume data (image) is generated, and Equation 2 is used to perform initial registration to minimize the average distance between the closed surfaces of the two volume data (image). (S131).

[Equation 2]

는 영상2의 점

로 변환된다.

는 영상 2에서 3차원으로 만들어진 챔퍼 거리맵의 점

의 위치에서의 상기 거리 값이다.

은 영상1과 영상2의 중첩된 영역에서의 점들의 개수이며, 각 점들에 대한 거리를 합산한 결과를

으로 나눔에 따라, 상기 폐 표면들 사이의 평균 거리가 산출된다.Here, each volume data will be described as an image 1,2, as follows. Dot of Image 1

Is the point in image 2

Is converted to.

Is the point on the chamfer distance map created in 3D in image 2.

The distance value at the position of.

Is the number of points in the overlapped area of Image 1 and Image 2, and the result of summing the distances for each point

By dividing by, the average distance between the lung surfaces is calculated.

,

,

)와, 3개의 회전 변환인자(

,

,

)와, 3개의 확대/축소 변환 인자(

,

,

)에 대한 변환식은 다음의 수학식 3과 같이 결정된다.The transform function uses an affine transform that adds a zoom transform to a rigid transform function consisting of a shift transform and a rotation transform. Three shift transform factors that are rigid

,

,

) And three rotation conversion factors (

,

,

) And three zoom conversion factors (

,

,

The conversion equation for) is determined as in Equation 3 below.

&Quot; (3) "

와

는 영상1에서의 복셀 및 중심의 좌표이고,

와

는 영상2에서의 복셀과 중심의 좌표이다. 변환 인자의 탐색을 위하여, 파웰의 방향 기법을 적용하여 빠른 시간 내에 최적의 위치로 수렴하도록 한다. 물론,

는 회전 각도를 의미한다.

Wow

Is the coordinate of the voxel and the center of the image 1,

Wow

Is the coordinate of the voxel and the center in the image 2. To search for the conversion factor, Powell's direction technique is applied to converge to the optimum position in a short time. sure,

Means the angle of rotation.

After the global motion compensation is completed using the affine matching technique, the control analyzer 140 corrects the local motion between the binary images using a local deformation model based non-rigid matching technique (S140).

의 근방

에서, 오차 함수를 최소화하는 변환 인자들을 수학 식 4를 이용하여 각 점마다 계산하게 되면, 최적의 비강체 정합이 가능하다.First, one point

Near

In Equation 4, the optimal non-rigid registration is possible by calculating the conversion factors for minimizing the error function for each point using Equation 4.

&Quot; (4) "

는 점

에 대한 선형 어파인 변환의 인자들이고,

는 점

에 대한 이동 변환의 인자들이다. 점

를 중심으로 근방

는 5×5×5 정육면체로 근사한다. 변환 인자들이 공간적으로 조금씩 변화한다고 가정하며, 상기 변환 인자들의 공간적 변화를 최소화하기 위해, 수학식 5의 평활화(smoothness) 항을 오차 함수에 추가할 필요가 있다.In Equation 4,

Point

Are factors of a linear affine transformation for,

Point

Are the arguments to the shift transform for. point

Near

Approximates to a 5 × 5 × 5 cube. It is assumed that the transformation factors change little by little in space, and in order to minimize the spatial variation of the transformation factors, it is necessary to add the smoothness term of Equation 5 to the error function.

[Equation 5]

Eventually, the optimal transformation factor, which is locally transformed and smoothly changes globally, is calculated to minimize the error function of Equation 6 below, which satisfies Equations 4 to 5 simultaneously.

&Quot; (6) "

The optimization of the error function is performed with the Newton-Raphson iterative technique. The moving image in each repetition is warped according to the optimal local transformation currently obtained, so that the next repetition calculation calculates the optimal conversion factors in the warped image and the fixed image. In addition, a Gaussian pyramid is generated for each of two images to be matched, and a multi-resolution technique is applied. An example of a 3D volume rendering result of the minimally and maximally breathing images that have undergone such registration is shown in FIG. 5.

On the other hand, the similarity measurement value for each matching result is compared with the error boundary (Error Bound), if the similarity is less than the error boundary proceeds to step (S150) thereafter, if the error boundary or more the affine matching process The operation of repeating the matching process by retrying operation S130 may be further added.

After performing each matching technique, the control analysis unit 140 generates a subtracted image between the binary images whose correspondence is identified through the global and local motion correction (S150), and analyzes the subtracted image. Quantification by (S160). Here, the subtracted image means an image corresponding to a brightness value obtained by subtracting the brightness value of the binary image for the maximum breath from the brightness value of the binary image for the minimum breath.

Next, the control analyzer 140 visualizes the quantified subtracted image on the screen through the display unit 120 (S170). That is, referring to FIG. 6A, the information about the subtracted image may be overlaid on the lung parenchyma of the existing maximum respiratory CT image.

On the other hand, in the black and white image of Figure 6 (a) has a disadvantage that it is not easy to determine the difference between the existing image and the overlayed image. Therefore, referring to FIG. 7B, the control analyzer 140 colorizes the brightness value of the subtracted image, color-maps the result to the CT image of the maximum breath or the minimum breath, and displays the display unit ( 120). Here, a rainbow color table may be used for color mapping of the subtracted image information. For example, in FIG. 7B, the portion of the blue system means a portion where the increase in CT brightness is small between the minimum breath CT image and the maximum breath CT image, and these portions represent the air trapping region.

The above-mentioned automatic air capture quantification method is summarized as follows. Each binary image is generated from the minimum / maximum breathing chest volume data obtained by the CT technique (S110 to S120). In addition, by using the affine matching method to minimize the distance between the lung surface of the minimum / maximum breath and the non-rigid registration method based on the regional deformation model, the global and local movement between the two binary images is corrected (S130 ~ S140). Then, after aligning and subtracting each binary image to generate a subtracted image, it is analyzed to quantify and visualize air capture information (S150 to S170).

On the other hand, the automatic air capture quantification method using the CT image as described above, it can be implemented as a computer-readable code on a computer-readable recording medium. Computer-readable recording media include all types of recording devices on which data is stored as media that can be read and executed by a computer system.

Examples of computer-readable recording media include ROM, RAM, CO-ROM, magnetic tape, floppy disks, optical data storage devices, and the like, which are also implemented in the form of carrier waves (for example, transmission over the Internet). Include. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

The automatic air capture quantification method as described above can visually and quantitatively diagnose the air trapping area of the lung disease patient by using the CT images of the lung disease patients. Applicable as a tool. In other words, the change in the brightness value of the lung parenchyma can be automatically analyzed in the minimally respiratory pulmonary CT images and the maximum respiratory pulmonary CT images.

Furthermore, if a medical form is changed by automating a qualitative reading method which is currently dependent on a doctor's manual operation and verifying it, more quantitative and sophisticated reading may be possible. In addition, if the automatic air capture quantification method is applied to the existing three-dimensional CT device or a picture archive and communication system, a significant amount of software sales and exports can be obtained. Furthermore, it will be able to strengthen the competitiveness of domestic medical systems in the global market.

Although the present invention has been described with reference to the embodiments shown in the drawings, this is merely exemplary, and it will be understood by those skilled in the art that various modifications and equivalent other embodiments are possible. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

1 is a flow chart of an automatic air capture quantification method using a chest CT image according to an embodiment of the present invention,

2 is another flow chart of FIG. 1;

3 is a system diagram for the method of FIG. 1;

4 is an exemplary diagram of a generated binary image;

5 is an exemplary diagram of a 3D volume rendering result of a minimum / maximum breathing image that has undergone a matching process;

6 is an exemplary diagram of an air capture visualization result through color mapping of a subtracted image and a subtracted image generated after registration of FIG. 5.

<Brief description of symbols for the main parts of the drawings>

100: automatic air capture quantification system 110: the imaging unit

120: display unit 130: input unit

140: control analysis unit

Claims (6)

(a) acquiring three-dimensional minimum breathing chest volume data and three-dimensional maximum breathing chest volume data through a CT technique; (b) dividing the lung parenchyma and the pulmonary vessel from the three-dimensional minimum breathing chest volume data and the three-dimensional maximum breathing chest volume data, and generating a binary image for each volume data; (c) correcting global movement between binary images for each of the volume data by using an affine matching technique for minimizing the distance between the closed surfaces of the minimum and maximum breaths; (d) correcting local motion between the binary images using a regional deformation model based non-rigid registration technique; (e) generating a subtracted image between the binary images whose correspondence is identified through the global and local motion correction; And (f) analyzing and quantifying the subtracted image, automatic air capture quantification method using a chest CT image. The method according to claim 1, wherein step (b), The lung parenchyma is a value of '0', the pulmonary blood vessels and the outside of the lungs are assigned a value of '1' to generate the binary image, automatic air capture quantification method using a chest CT image. The method according to claim 1, after the step (f), (g) further comprising visualizing the quantified subtracted image on a screen, automatic air capture quantification method using a chest CT image. The method of claim 3, wherein the subtracted video, And an image corresponding to the brightness value obtained by subtracting the brightness value of the binary image for the maximum breath from the brightness value of the binary image for the minimal breathing. The method according to claim 4, wherein (g) step, And colorizing the brightness value of the subtracted image and visually mapping the color value to the CT image of the maximum breath or the minimum breath. A computer-readable recording medium having recorded thereon a program for executing the method of any one of claims 1 to 5 on a computer.

Images